What is the job of autogen.sh when building a c++ package on Linux [duplicate]

Question

I saw a common pattern when installing a c/c++ package from source on Linux (Ubuntu 16.04):

1. ./autogen.sh
2. ./configure
3. make
4. make install

I understand make and make install, and I guess configure creates a Makefile based on user preferences, but I don't see why autogen.sh is necessary.

Does anyone know what it is there for?

Answer 1

The steps:

1. The autogen.sh script generates the configure script (from configure.ac, using autoconf) and any files it needs (like creating Makefile.in from Makefile.am using automake). This requires autotools to be installed on your system, and it must be run when checking out the project from source control (if configure isn’t checked in). People who download source tarballs can usually skip this step, because output of this step is included in source tarballs.

   Note This is usually equivalent to autoreconf --install. If there is not autogen.sh file, then just run autoreconf --install instead. If you have inherited a project with an autogen.sh, consider deleting it if you can use autoreconf --install.

2. The configure script generates Makefile and other files needed to build. Typically Makefile.in is used as a template to generate Makefile (and config.h.in to generate config.h). This process happens using only standard tools installed on your system, like sed and awk, and doesn't require autotools to be installed.

3. The make command builds the software.

4. The make install command installs it.

These are broken into different steps because they are often run at different times. The autogen.sh step is traditionally run by people who are developing the software, since they are expected to install autoconf on their systems and they make changes to configure.ac. End-users are not expected to have autotools installed.

These expectations have been changed a bit now that end-users are more likely to check a project out of source control instead of downloading source releases.

Answer 2

This applies only to programs / libraries, which are built using the autotools build chain. It generates the files, which are configured by the configure script. The configure script then populates .in files and generates Makefiles from Makefile.am templates. Which can finally be used to compile, link and install the program / library.

It's becoming slowly obsolete with the move to multi platform packages. CMake and more modern tool chains are state of the art.